Hello, I've been testing VPPs NAT64 plugin in 25.06 and 26.02 on a box with ixgbe NICs the last days and have ran into some issues.
The first issue I had was that I'd see quite high RTT(~4 times what I'd see over native IPv4 for the same IP without translation) when using certain worker configurations > 1. 10 workers would work, for example, but 11 would not. This also seems to be alleviated when using num-rx-queues set >1. The second issue I was seeing was that even though I set up a big pool, only the last IP in the range would be used. For example "nat64 add pool address 10.0.0.2 - 10.0.0.128" would only result in sessions being put on 10.0.0.128 with all the other IPs having no sessions, even as the last IP in the pool approached 65535 ports used. When all the outside ports were filled up new sessions were being created very slowly at best, I assume they were being recycled. I realize I'm light on details right now as I haven't had time to dig too much into the source code for the plugin but I was wondering if anyone has any theories on what could be wrong or how to troubleshoot it further. Regards, Wilhelm
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#27122): https://lists.fd.io/g/vpp-dev/message/27122 Mute This Topic: https://lists.fd.io/mt/120470935/21656 Group Owner: [email protected] Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
